Breast Cancer Genome-Wide Association Studies: There Is Strength in Numbers

D Fanale1,3, V Amodeo1,3, LR Corsini1,3, S Rizzo1, V Bazan1,2 and A Russo1,2

1Department of Surgical and Oncological Sciences, Section of Medical Oncology, University of Palermo, Palermo, Italy and 2Sbarro Institute for Cancer Research and Molecular Medicine, Center for Biotechnology, College of Science and Technology, Temple University, Philadelphia, PA, USA

Breast cancer (BC) is a heterogeneous disease that Introduction exhibits familial aggregation. Family linkage studies have identified high-penetrance genes, BRCA1, BRCA2, Breast cancer (BC) is the most common cancer and the PTEN and TP53, that are responsible for inherited BC second leading cause of cancer death among women syndromes. Moreover, a combination of family-based and (Parkin et al., 2005). population-based approaches indicated that genes involved The family history is the main risk factor for BC, in DNA repair, such as CHEK2, ATM, BRIP and indicating that the genetic factors are very important in the PALB2, are associated with moderate risk. Therefore, all development of disease (Antoniou and Easton, 2006). of these known genes account for only 25% of the familial In the 1990s, linkage studies in multiple case families have aggregation cases. Recently, genome wide association identified two major susceptibility genes in BC BRCA1 and studies (GWAS) in BC revealed single nucleotide poly- BRCA2 (Miki et al., 1994; Wooster et al., 1995). morphisms (SNPs) in five novel genes associated to Germline mutations in BRCA1 and BRCA2 genes susceptibility: TNRC9, FGFR2, MAP3K1, H19 and occur rarely in the general population but confer high lymphocyte-specific protein 1 (LSP1). The most strongly risks of breast and ovarian cancer and a lower risk for associated SNP was in intron 2 of the FGFR2 gene that is other cancers (Antoniou et al., 2003; Thompson and amplified and overexpressed in 5–10% of BC. rs3803662 Easton, 2004). of TNRC9 gene has been shown to be the SNP with the TP53 and PTEN mutations are also present in the strongest association with BC, in particular, this poly- population at low frequency and lead to very high BC morphism seems to be correlated with bone metastases risk associated with rare cancer syndrome, however, and estrogen receptor positivity. Relevant data indicate population-based studies have estimated that alterations that SNP rs889312 in MAP3K1 is correlated with BC in these genes account only the 15% of the familial risk susceptibility only in BRCA2 mutation carriers, but is not of BC (Sidransky et al., 1992; FitzGerald et al., 1998; associated with an increased risk in BRCA1 carriers. Peto et al., 1999; Dite et al., 2003). Finally, different SNPs in LSP1 and H19 and in minor Further, genetic linkage analyses failed to identify genes probably were associated with BC risk. New additional high-penetrance susceptibility genes and the susceptibility allelic variants associated with BC risk were identification of rare variants of genes involved in DNA recently discovered including potential causative genes repair, such as CHEK2, ATM, BRIP and PALB2 in involved in regulation of cell cycle, apoptosis, metabolism families lacking BRCA mutations (Meijers-Heijboer and mitochondrial functions. In conclusion, the identifica- et al., 2002; Thompson et al., 2005; Rahman et al., tion of disease susceptibility loci may lead to a better 2007; Hollestelle et al., 2010), associated with a understanding of the biological mechanism for BC to moderate risk of disease, can explain only a small improve prevention, early detection and treatment. portion of familial risk. Oncogene (2012) 31, 2121–2128; doi:10.1038/onc.2011.408; Therefore, all of these known genes account for only published online 26 September 2011 25% of the familial aggregation cases (Thompson and Easton, 2004), suggesting that most of the familial risk Keywords: FGFR2; GWAS; H19; LSP1; MAP3K1; of BC can plausibly involve a combination of multiple TNRC9 low-penetrance susceptibility alleles, each conferring a small effect on BC risk (Antoniou and Easton, 2006; Table 1). According to this model defined ‘polygenic’, proposed to explain the genetic susceptibility to BC, a large number of low-risk variants occurs with high frequency Correspondence: Professor A Russo, Department of Surgical and in populations, therefore, it may have a multiplicative Oncological Sciences, Section of Medical Oncology, Universita` di effect in determining the overall risk of disease (Pharoah Palermo, via del Vespro 129, 90127 Palermo, Italy. et al., 2002; Figure 1). A significant part of polygenic E-mail: [email protected] 3These authors contributed equally to this work contribute to low-penetrance susceptibility may rise by Received 10 July 2011; revised 9 August 2011; accepted 9 August 2011; non-conservative missense mutations in evolutionarily published online 26 September 2011 conserved domains. GWAS in breast cancer D Fanale et al 2122 Table 1 Genetic loci implicated in breast cancer susceptibility population the identification of a susceptibility allele with a frequency of 5% requires over 800 cases. In the High penetrance, Low penetrance, Low penetrance, low frequency low frequency high frequency same population, the identification of a susceptibility allele with a frequency of 1% requires over 3700 BRCA1 CHEK2 FGFR2 unselected cases, whereas about 700 would be enough BRCA2 ATM TNRC9 if three affected families are selected. Therefore, the p53 PALB2 LSP1 PTEN BRIP1 MAP3K1 power of association studies can be significantly SLC4A7 increased using selected cases with a family history of COX11 cancer because less cases are required to demonstrate the association with disease (Houlston and Peto, 2004). The potential of the association studies of cases with a family history to identify low-penetrance alleles conferring a relative risk of 2 has been demonstrated by the mutation CHEK2 1100delC in patients with BC. This variant carried by 1% of the population confers an increased risk of 1.7-fold. The frequency was not significantly increased in unselected cases (1.4%), but it was strongly increased in familial cases without BRCA1 and BRCA2 mutations (5.1%; Meijers-Heijboer et al., 2002). In the past years several novel risk alleles for BC were identified by four recent GWA studies: Breast Cancer Association Consortium, Cancer Genetic Markers of Susceptibility, DeCode Islanda, Memorial Sloan–Ket- tering Cancer Center (Easton et al., 2007; Hunter et al., 2007; Gold et al., 2008; Stacey et al., 2008). In each of them the association study was shared into Figure 1 Allelic variants at low frequency tend to be associated with higher relative risk of BC (for example, BRCA1, BRCA2), three phases: the first phase identifies the common SNPs high-frequency allelic variants are associated with lower RR in cases and controls, the second phase evaluates how (for example, FGFR2, LSP1, and so on) configuring an inverse many of the above SNPs are common to a greater correlation. number of cases and controls and, finally, the third phase aims to identify new alleles of susceptibility of BC. Easton et al., in their study, identified five indepen- Genome-wide association studies (GWAS) dent loci associated with increased susceptibility to BC (Po10À7). This multistage study involved in the first In recent years the research of low-penetrance allelic stage 390 BC cases with a strong family history and variants was conducted mainly through GWAS. These 364 controls, and 3990 cases and 3916 controls in the studies use a large number of common genetic single second stage. nucleotide polymorphisms (SNPs) to identify associa- To define the risk associated with the 30 most tions with disease that rely upon patterns of linkage significant SNPs, a third stage of the study was disequilibrium (LD) in the human genome (Hirshfield conducted involving 21 860 cases and 22 578 controls et al., 2010). The power of GWAS is to evaluate the from 22 additional studies in the Breast Cancer association of genetic variants at different loci on Association Consortium. different chromosomes (LD) in large series of cases These combined analyses allowed to observe that the versus controls, analyzing a panel of hundred thousand SNPs showing a stronger statistical evidence of associa- SNPs simultaneously, to identify new alleles of suscept- tion with an increased familial risk were: rs2981582 lies ibility to BC (Orr and Chanock, 2008). In the human in intron 2 of FGFR2, rs12443621 and rs8051542 within genome has been estimated that there are seven million TNRC9, rs889312 lies in a region that contain MAP3K1 of common SNPs that have a minor allele frequency gene, rs3817198 lies in intron 10 of lymphocyte-specific (m.a.f.), 45% and because recombination occurs in protein 1 (LSP1) and rs2107425 within the H19 gene. different hot-spots, the nascent polymorphisms are often Unlike other BC susceptibility genes previously strongly correlated. identified that are involved in DNA repair and sex These studies therefore provide a powerful tool to hormone synthesis, in this work three of the five loci identify novel markers for susceptibility and prognosis reported contain genes involved in regulation of cell of disease (Peto, 2002; Houlston and Peto, 2004; Easton growth and cell signaling (Easton et al., 2007). and Eeles, 2008). In the GWA studies the accumulation Starting from this study of Easton et al., in attempt of a large number of data is crucial. Houlston and Peto, to identify further loci associated with BC risk, (2004) have estimated the number of cases required to Ahmed et al. have genotyped in a third stage further identify low-penetrance alleles conferring a relative risk 814 SNPs, involving 3878 cases and 3928 controls from of two both in an unselected population and in families three studies of the Cancer Genetic Markers of with first-degree relatives affected. In an unselected Susceptibility.

Oncogene GWAS in breast cancer D Fanale et al 2123 These analyses allowed to identify three additional actylidin, that functions as a ubiquitin protein ligase SNPs (rs4973768, rs4132417 and rs6504950) that have (E3; Mani and Gelmann, 2005), that could have a role in been evaluated in a fourth stage from 27 studies in the breast tumorigenesis. Breast Cancer Association Consortium. To identify new risk variants associated with BC rs4132417 showed no evidence of association in the susceptibility, Stacey et al. have carried out a GWA fourth stage, it was probably a false positive. rs4973768 study genotyping 1600 Icelandic individuals with BC showed strong association with age that was higher for and 11 563 controls. ER-positive than ER-negative disease, moreover there Two SNPs showed statistically significant association was no evidence of association with a positive family with BC: the A allele of rs13387042 on chromosome history of BC. 2q35 and the T allele of rs3803662 on 16q12. Similarly, rs6504950 showed statistical evidence of The 25% of individuals of European descent are association with ER-positive disease and no association homozygous for allele A of rs13387042 and have an with the family history, and unlike of rs4973768 no estimated 1.44-fold greater risk than noncarriers and association with the age. about 7% are homozygous for allele T of rs3803662 and Moreover, in this study additional association ana- have a 1.64-fold greater risk. lyses showed that another SNP rs1357245 is located in These risk alleles were not associated with histo- the same LD block as rs4973768 in the 3p24 region pathological subtype, stage and grade of tumors, but (Ahmed et al., 2009). Genotyping the 28 SNPs, known confer preferential risk for estrogen receptor (ER)- to be present in this region, in 2301 cases and 2256 positive BC. controls, it presents a further SNP rs2307032 that was In the LD block where lies rs13387042 there are no correlated with both SNPs. known genes, but there are proximally and distally Hunter et al., in a recent work, have identified alleles TNP1, IGFBP5, IGFBP2 and TNS1 genes (Stacey et al., in FGFR2 associated with risk of sporadic postmeno- 2007). pausal BC. rs3803662 is near the 50 end of TNRC9, whose In this study, the National Cancer Institute Cancer increased expression is highly predictive of metastasis Genetic Markers of Susceptibility identified four SNPs, to bone of BC (Smid et al., 2006). two (rs1219648 and rs2420946) in intron 2 of FGFR2 Comparing the results obtained from four major and two (rs11200014 and rs28981579) in FGFR2 gene, studies of GWA, it has been highlighted a correlation of using GWAS of BC by genotyping 528 173 SNPs in 1145 allele frequency of some SNPs located on the genes: postmenopausal women of European ancestry with FGFR2 encoding a receptor tyrosine kinase, TNRC9 invasive BC and 1142 controls. encoding a high-mobility group chromatin-associated These polymorphic variants showed a strong associa- protein, MAP3K1, which encodes the signaling protein tion with the risk of disease (Hunter et al., 2007). mitogen-activated protein kinase 1 (MAPK1), LSP1 A similar experimental design was conducted by Gold encoding LSP1 and H19 an untranslated messenger et al. in 249 Ashkenazi Jewish women, containing multiple RNA involved in regulation of the insulin growth factor cases of BC but lacking BRCA1 or BRCA2 mutations, gene 2 (Table 2). presented at the Memorial Sloan–Kettering Cancer Center. These new discovered susceptibility genes are differ- This study confirmed the BC association with the entially expressed between the five distinct molecular FGFR2 locus, identified by Easton and Hunter studies, subtypes of BC, based on differential gene expression but showed an association with the RNF146 and profiles: luminal A, luminal B, basal like, ErbB2 þ and ECHDC1 region at 6q22.33 not seen in the previous normal like (Sorlie et al., 2003). These distinct molecular works (Gold et al., 2008). subtypes of BC are associated with different clinical ECHDC1 gene encodes for a trifunctional protein outcomes (Sorlie et al., 2001). involved in mitochondrial fatty acid oxidation (Hashimoto If the probability to develop a given subtype of BC is et al., 1996) and RNF146 encodes for a protein, called genetically determined, we would expect to find that the

Table 2 Comparative analysis of the SNPs identiﬁed in the four studies (BCAC, CGEMS, MSKCC and DeCode Islanda) Gene Location BCAC CGEMS MSKCC DeCode Islanda

FGFR2 Chr 10q rs2981582 rs1219648 rs2981582 No (in intron2) rs2420946 (in intron2) rs11200014 and rs28981579 TNRC9 Chr 16q rs12443621 rs8049226 (within 200 kb of TNRC9) rs3803662 and rs3112625 rs3803662 (near TNRC9) rs8051542 MAP3K1 Chr 5q rs889312 rs726501 No No LSP1 Chr 11p rs3817198 rs7120258 No No rs498337 H19 Chr 11p rs2107425 rs7120258 No No rs7578974

Abbreviation: BCAC, Breast Cancer Association Consortium; CGEMS, Cancer Genetic Markers of Susceptibility; MSKCC, Memorial Sloan– Kettering Cancer Center; SNPs, single nucleotide polymorphisms.

Oncogene GWAS in breast cancer D Fanale et al 2124 newly discovered susceptibility genes (Easton et al., 2q35 and have an estimated 1.44-fold greater risk than 2007) are differentially expressed in the various tumor noncarriers. Risk from both alleles was confined to subtypes. ER-positive tumors. Recently a significantly differential mRNA expression The variant in the 5p12 region, which is close to the of TNRC9, FGFR2, MAP3K1, H19 and LSP1 from 112 FGFR2 ligand FGF10, also shows strong evidence of an breast tumor samples, representing all five subtypes, has association primarily with ER þ tumors (Stacey et al., been identified by analysis of variance (Nordgard et al., 2008). 2007). These data show the necessity to conduct stratified SNP disease association studies and to select patients by their molecular subtypes, to confer more power to the TNRC9 GWA studies. The locus on 16q includes a gene TNRC9 and a hypothetical gene LOC643714. TNCR9 (also known as TOX3) is a gene of uncertain function containing a tri- FGFR2 nucleotide repeat motif and encoding a member of the high-mobility group family of non-histone chromatin FGFR2 is a member of a receptor tyrosine kinase gene proteins. The presence of a putative high-mobility group superfamily, which contributes to the cell growth, box motif suggests that it might function as a transcrip- invasiveness, motility and angiogenesis (Ricol et al., tion factor (Easton et al., 2007). Several studies have 1999). Overexpression of FGFR2, one of the common shown that susceptibility loci at TNCR9 predispose to low-penetrance susceptibility genes, is observed in breast sporadic BC. Rs3803662, located near the 50 end of tumor tissues (Adnane et al., 1991) and in BC cell lines TNRC9, has been shown to be the SNP with the (Tannheimer et al., 2000). Its expression is associate strongest association with BC. The SNP rs3803662 is with ER þ tumors (Luqmani et al., 1992), suggesting a related to both ER þ and ERÀ tumors (McInerney hormone-dependent action of this gene. Recently, gene et al., 2009). expression studies have shown increasing FGFR2 The associations of rs3803662 with other SNPs seem expression levels associated with the rare homozygote to be not significant. Other two SNPs (rs12443621 and genotype and functional studies identified the OCT1/ rs8051542) with important evidence of association are RUNX2-binding site as the main determinant of the located in an LD block containing the 50 end of TNRC9. increased expression levels (Meyer et al., 2008). Aber- Furthermore, Hunter et al. (2007) showed that there is rant expression of alternatively spliced isoforms of only one SNP significant (rs8049226) within 200 kb of FGFR2 transforms BC cells by sustained signal TNRC9. In contrast, the coding region of TNCR9 transduction (Moffa and Ethier, 2007). The FGFR2 contains SNPs showing no evidence of association. The gene, located at chromosome 10q26, contains at least 22 A allele of rs13387042 located on chromosome 2q35 (A- exons (Ingersoll et al., 2001). rs13387042) and the T allele of rs3803662 on 16q12 (T- Several mutations and common SNPs within or rs3803662) confer increased risk of BC for ER-positive flanking the FGFR2 gene have been identified. A tumors. Any interaction was observed between the number of studies have been conducted to investigate 2q35 and 16q12 loci. Moreover, no known gene or the association between FGFR2 polymorphisms and the human RNA was found for the LD block containing risk of BC in humans. The association is restricted to rs13387042. The rs3803662 SNP located on 16q12 SNPs in the LD block covering intron 2. In particular, occurs in the fourth exon of a poorly characterized three polymorphic variants, rs1219648 (A4G), mRNA. In BC, the q arm of chromosome 16 is rs2420946 (C4T) and rs2981582 (C4T) are more frequently lost, therefore, it is likely that one or more investigated for their closed correlation with BC. Easton tumor suppressor genes are present in the same region. et al. (2007) showed that rs2981582 had a clear relevance Differences in stage, grade or histopathological subtype to BC. were not significantly correlated with the low-penetrance Gold et al. (2008) confirmed the previously reported susceptibility alleles, and there was no significant results for FGFR2 locus. Recently, a further GWAS difference in allele frequencies between in situ and study confirmed the correspondence between FGFR2 invasive carcinoma. In African Americans, T allele of susceptibility loci and BC risk. In particular, the per- the SNP rs3803662 was significantly protective and, allele odds ratio was higher for ER-positive rather than thus, it was not associated with increased BC risk for ER-negative BC (Ahmed et al., 2009). This finding is (Stacey et al., 2007). Three susceptibility alleles consistent with the involvement of FGFR2 in estrogen- (rs2981582, rs3803662 and rs13281615) also have shown related breast carcinogenesis (Tamaru et al., 2004), and an evidence of association with family history of BC. In with higher levels of FGFR2 expression in ER þ than fact, each of these SNPs was more frequent in women ERÀ cell lines and tumors (Zhang et al., 1999). Stacey with a first-degree relative with the disease than in those et al. genotyped B300 000 SNPs in 1600 Icelandic without. Furthermore, an evidence of association with individuals with BC and 11 563 controls. They found breast in situ carcinoma has been shown by three SNPs that 25% of individuals of European descent are (rs2981582, rs3803662 and rs889312; Easton et al., homozygous for allele A of rs13387042 on chromosome 2007). Increased expression of TNRC9 indicates a major

Oncogene GWAS in breast cancer D Fanale et al 2125 susceptibility to metastasis of BC to bone. ER positivity associated with risk of developing non-Hodgkin’s is predictive of bone metastases. The possible effects of lymphoma (Cerhan et al., 2007). The most important the correlation between rs3803662, TNRC9, bone GWASs reported different conclusion about the role of metastases and ER positivity remain to be explicated. LSP1 gene in BC susceptibility. Easton et al. (2007) Many association studies have shown that SNPs in reported one SNP (rs3817198) lies in intron 10 of LSP1 FGFR2, TNRC9 and MAP3K1 increase the BC risk in gene with P values in the range 10À5–10À9; Gold et al. BRCA2 mutation carriers with a similar relative risk to (2008) found two SNPs (rs3817198, rs498337), near the that seen in the general population. In contrast, in LSP1 region, with P values in the range of Po0.01 by BRCA1 mutation carriers only the rs3803662 SNP was allele test, where Hunter et al. (2007) provided evidence associated with an increased BC risk (Easton and Eeles, for one SNP (rs7120258) in the region with a P value 2008). 0.01. Recent study identified that LSP1 minor allele of rs3817198 was associated with increased BC risk only for BRCA2 mutation carriers (Antoniou et al., 2009). A further SNP, rs2107425, located just 110 kb from MAP3K1 rs3817198, was also identified (overall P ¼ 0.00002). rs2107425 is within the H19 gene, an imprinted MAP3K1 (MEKK1) encodes the MAPK protein that maternally expressed untranslated messenger RNA phosphorylates and activates the MAPK kinase closely involved in regulation of the insulin growth (MAPK2) that in turn phosphorylates the MAPK/ factor gene 2 (Easton et al., 2007). In H19 region on ERK to produce downstream signaling effects on a chromosome 11p, Easton et al. (2007) reported P values variety of cancer genes. MAP3K1 forms part of the in the range 0.01–10À5, Gold et al. saw no signal, MAPK cell signaling pathway implicated in cellular whereas Hunter et al. (2007) found two SNPs response to mitogens. The MAPK pathway is strongly (rs7120258, rs7578974), with association P values in linked to HER2 receptor activity and activating muta- the range of 0.01, with one additional SNP, rs217228, tions in the MAPK pathway have been associated with with a P value in the range of 0.02. HER2 þ breast tumors (Bild et al., 2006; Creighton et al., 2006). MAP3K1 was identified by Easton et al. (2007) to have a per-allele odds ratio effect of 1.13 (95% Recently discovered BC susceptibility loci confidence interval: 1.09–1.18). MAP3K1 effects were found to be relevant in ER þ and PR þ tumors to a New susceptibility allelic variants associated with BC greater degree than in ERÀ or PRÀ tumors (Garcia- risk were recently discovered through large replication Closas and Chanock, 2008). MAP3K1 is differentially studies in combination with the original GWAS data. expressed in different BC subtypes (Nordgard et al., The combination between these studies and GWAS data 2007). Hunter et al. (2007) found only one SNP allowed to identify three SNPs: rs4973768, rs4132417 (rs726501) with a P value in the range of P 0.01 by o and rs6504950. There is a strong evidence for additional allele test. Gold in the Memorial Sloan–Kettering susceptibility loci on 3p and 17q. The region 3p24 Cancer Center study and Stacey, in the DeCode study, includes two potential causative genes, SLC4A7 and did not see significant SNPs between individuals with NEK10. SLC4A7 (solute carrier family 4, sodium BC and controls. GWA studies conducted by Garcia- carbonate cotransporter, member 7) is a potential Closas and Chanock identified that the rs889312 variant tyrosine kinase protein whose expression is reduced in is in a LD block containing the MAP3K (Garcia-Closas BC specimens and cell lines. NEK10 (never-in mitosis- et al., 2008). related kinase 10) belongs to a family of 11 never in The Consortium of Investigators of Modifiers of mitosis a-related kinases that are involved in cell cycle BRCA1/2 has recently evaluated whether variants in regulation. However, unlike other NEKs, no role has FGFR2 (rs2981582), TNRC9 (rs3803662) and MAP3K1 been associated to NEK10. A 300-kb LD block on (rs889312) are associated with the risk of BC in over 17q23.2 includes rs6504950 that lies in intron 1 of 10 000 BRCA1 and BRCA2 mutation carriers from STXBP4 (syntaxin-binding protein 4), codifying an 23 studies (Antoniou et al., 2008). The evidence of insulin-regulated STX4-binding protein involved in the association with SNP rs889312 in MAP3K1 was weaker regulation of GLUT4 vesicle translocation and glucose and was restricted to BRCA2 mutation carriers, how- transport. The rs6504950 allelic variant showed no ever, this SNP was not associated with an increased risk association with age or family history, but a stronger in BRCA1 carriers. association in ER-positive disease versus ER-negative disease. The same LD block includes other genes as COX11 (cytocrome C assembly protein 11), that is LSP1 and H19 located 10 kb upstream of rs6504950, and TOM1L1 (target of myb-1-like1). In lymphocytes, the rs6504950 LSP1 gene (also known as WP43) encodes an F-actin risk allele is correlated with higher expression of COX11 bundling cytoskeletal protein expressed in hematopoie- levels, but no association has been shown with expres- tic and endothelial cells. LSP1 has been implicated in sion levels of either TOM1L1 or STXBP4. Allele malignant lymphoma and Hodgkin’s disease (Marafioti frequency studies in European populations have et al., 2003), and other variants in this gene have been revealed that rs4973768 and rs6504950 could explain

Oncogene GWAS in breast cancer D Fanale et al 2126 respectively 0.4% and 0.07% of the familial risk of BC. rs3803662 (16q12), rs889312 (5q11), rs3817198 (11p15) These susceptibility loci together with those previously and rs13387042 (2q35) but only two of them (16q12 and identified in original GWAS would give rise to 5.9% of 2q35) were associated with tumors with the core basal BC familial risk (Ahmed et al., 2009). Further genome- phenotype (Pp0.002). This study identifying novel risk wide linkage studies have revealed three putative BC factors associated with BC subtypes could allow a better susceptibility regions of interest, located on 3q25, 6q24 tumor stratification resulting in prevention, early detec- and 21q22. Moreover, it has been observed that the tion and treatment improvement (Broeks et al., 2011). allelic variants on both chromosomes 21 and 3 were correlated with a higher percentage of bilaterality and a higher number of familial cases (Rosa-Rosa et al., 2009). The power of GWAS A recent GWAS has identified a novel polymorphic variant rs11249433 within the 1p11.2 region associated The GWAS represents a new powerful approach to with BC risk. identify lower penetrance alleles that cannot be detected This association is stronger in ER-positive than by genetic linkage studies. The risk conferred by these ER-negative tumors, is correlated with mRNA expres- alleles individually is too weak, generally 1.3-fold or less, sion of NOTCH2 gene and is highest in breast tumors but the combined effects may be useful for risk without TP53 mutations. prediction (Easton et al., 2007). This would promote Further studies are needed to evaluate the possible the development of novel methodologies for analysis of role of rs11249433 in NOTCH2 regulation and BC data generated by large-scale SNP studies. In recent development (Fu et al., 2010). Other variants can years, the research and identification of low-penetrance significantly modify the BC risk in BRCA1 and BRCA2 susceptibility loci played a key role in the etiology of BC mutations carriers. The rs6138178 in SNRPB and and, in particular, of those BCs that have estrogen and rs6602595 in CAMK1D show a strongest association progesterone receptors. The combination of BC suscept- in BRCA1 carriers, whereas rs9393597 in LOC134997 ibility alleles together with other risk factors may be and rs12652447 in FBXL7 in BRCA2 carriers. important clinically and it may explain an appreciable These loci appeared to interact multiplicatively for BC fraction of the genetic variance in BC risk. The risk in BRCA1/BRCA2 carriers, therefore, these SNPs identification of the causative variants can be extremely together with other genetic and environmental factors problematic but the use of GWAS from multiple may improve the BC risk assessment in these popula- populations with different patterns of LD can reduce tions (Wang et al., 2010). the difficulty of analysis. The power of GWAS may be Recently, a new study (Black Women’s Health Study) increased by enlarging the number of samples in both has been conducted in a population of African Amer- the cases and the controls, and by identifying clinical ican Women (886 BC cases versus 1089 controls) to and molecular subtypes (Kristensen and Borresen-Dale, identify genetic variants associated with risk of BC. As 2008). However, the GWA experiments need the effort in the original study (Stacey et al., 2008), it has been of several research groups to collect a sufficient number confirmed the strong association of rs4415084 on 5p12 of patients for large multistage studies and they require with overall risk and ER-positive tumors. No associa- large amounts of money. The allele frequency of the tion was observed for ER- and PR-negative tumors. variant and the risk conferred by it will determine the Other susceptibility allelic variants identified from number of cases to be genotyped. There is a common BWHS are rs6451770, rs12515012, rs13156930 and scepticism toward these new approaches because it is not rs16901937. A 21% increase in risk was associated with known the mechanism by which the novel allelic each copy of the rs16901937 G-allele. The closest gene to variants cause the susceptibility. Furthermore, some these regions is MRPS30, involved in apoptosis and differences were found between different studies. This encoding a component of the mitochondrial ribosome. could be due to population stratification, sample-size Moreover, MRPS30 is involved in a gene expression differences or genetic heterogeneity in the setting of profile that allows to discriminate ER-positive from ER- different genotyping platforms (Perlegen, Mountain negative tumors (Ruiz-Narvaez et al., 2010). Other BC View, CA, USA; Affymetrix, Santa Clara, CA, USA; susceptibility alleles could be identified through large- Illumina, San Diego, CA, USA) and different algo- scale replication studies in combination with previous rithms to filter data (Gold et al., 2008). GWAS. However, these analyses have still a limited power. Recently, a novel multi-SNP GWAS analysis method In a recent work, the BC risk association with eight called Pathways of Distinction Analysis was developed. susceptibility loci identified by GWAS was investigated This method uses GWAS data and pathway-gene and in relation to specific breast tumor subtypes. A strong gene-SNP associations to identify pathways that could association was identified between ER þ tumors and six permit the distinction of cases from controls. Therefore, of eight loci identified by GWAS: rs2981582 (10q26), relating a pathway with the disease risk, for the SNPs rs3803662 (16q12), rs13281615 (8q24), rs13387042 associated with a pathway, the cases will be similar to (2q35), rs4973768 (3p24) and rs6504950 (17q23). A other cases than to controls. This method provides a most strongly relation was observed between two new analytical tool that could enrich the power of candidate loci, CASP8 (rs1045485, rs17468277) and GWAS in BC risk prediction (Braun and Buetow, 2011). TGFB1 (rs1982073) and PR tumors. Four loci were In conclusion, the recently discovered data could open associated with triple negative tumors (Pp0.016): up new streets for basic research. In future, a new

Oncogene GWAS in breast cancer D Fanale et al 2127 generation of large-scale association studies, in combi- Conﬂict of interest nation with replication analyses and multiple scans could be able to identify many more loci. The authors declare no conﬂict of interest.

References

Adnane J, Gaudray P, Dionne CA, Crumley G, Jaye M, Schlessinger J associations with five susceptibility loci by clinical and pathological et al. (1991). BEK and FLG, two receptors to members of the FGF characteristics. PLoS Genet 4: e1000054. family, are amplified in subsets of human breast cancers. Oncogene Gold B, Kirchhoff T, Stefanov S, Lautenberger J, Viale A, Garber J 6: 659–663. et al. (2008). Genome-wide association study provides evidence for a Ahmed S, Thomas G, Ghoussaini M, Healey CS, Humphreys MK, breast cancer risk locus at 6q22.33. Proc Natl Acad Sci USA 105: Platte R et al. (2009). Newly discovered breast cancer susceptibility 4340–4345. loci on 3p24 and 17q23.2. Nat Genet 41: 585–590. Hashimoto T, Shindo Y, Souri M, Baldwin GS. (1996). A new Antoniou A, Pharoah PD, Narod S, Risch HA, Eyfjord JE, Hopper JL inhibitor of mitochondrial fatty acid oxidation. J Biochem 119: et al. (2003). Average risks of breast and ovarian cancer associated 1196–1201. with BRCA1 or BRCA2 mutations detected in case Series Hirshfield KM, Rebbeck TR, Levine AJ. (2010). Germline mutations unselected for family history: a combined analysis of 22 studies. and polymorphisms in the origins of cancers in women. J Oncol Am J Hum Genet 72: 1117–1130. 2010: 297671. Antoniou AC, Easton DF. (2006). Models of genetic susceptibility to Hollestelle A, Wasielewski M, Martens JW, Schutte M. (2010). breast cancer. Oncogene 25: 5898–5905. Discovering moderate-risk breast cancer susceptibility genes. Curr Antoniou AC, Sinilnikova OM, McGuffog L, Healey S, Nevanlinna Opin Genet Dev 20: 268–276. H, Heikkinen T et al. (2009). Common variants in LSP1, 2q35 and Houlston RS, Peto J. (2004). The search for low-penetrance cancer 8q24 and breast cancer risk for BRCA1 and BRCA2 mutation susceptibility alleles. Oncogene 23: 6471–6476. carriers. Hum Mol Genet 18: 4442–4456. Hunter DJ, Kraft P, Jacobs KB, Cox DG, Yeager M, Hankinson SE Antoniou AC, Spurdle AB, Sinilnikova OM, Healey S, Pooley KA, et al. (2007). A genome-wide association study identifies alleles in Schmutzler RK et al. (2008). Common breast cancer-predisposition FGFR2 associated with risk of sporadic postmenopausal breast alleles are associated with breast cancer risk in BRCA1 and BRCA2 cancer. Nat Genet 39: 870–874. mutation carriers. Am J Hum Genet 82: 937–948. Ingersoll RG, Paznekas WA, Tran AK, Scott AF, Jiang G, Jabs EW. Bild AH, Yao G, Chang JT, Wang Q, Potti A, Chasse D et al. (2006). (2001). Fibroblast growth factor receptor 2 (FGFR2): genomic Oncogenic pathway signatures in human cancers as a guide to sequence and variations. Cytogenet Cell Genet 94: 121–126. targeted therapies. Nature 439: 353–357. Kristensen VN, Borresen-Dale AL. (2008). SNPs associated with Braun R, Buetow K. (2011). Pathways of distinction analysis: a new molecular subtypes of breast cancer: on the usefulness of stratified technique for multi-SNP analysis of GWAS data. PLoS Genet 7: Genome-wide Association Studies (GWAS) in the identification of e1002101. novel susceptibility loci. Mol Oncol 2: 12–15. Broeks A, Schmidt MK, Sherman ME, Couch FJ, Hopper JL, Dite GS Luqmani YA, Graham M, Coombes RC. (1992). Expression of basic et al. (2011). Low penetrance breast cancer susceptibility loci fibroblast growth factor, FGFR1 and FGFR2 in normal and are associated with specific breast tumor subtypes: findings from malignant human breast, and comparison with other normal tissues. the Breast Cancer Association Consortium. Hum Mol Genet 20: Br J Cancer 66: 273–280. 3289–3303. Mani A, Gelmann EP. (2005). The ubiquitin-proteasome pathway and Cerhan JR, Ansell SM, Fredericksen ZS, Kay NE, Liebow M, Call TG its role in cancer. J Clin Oncol 23: 4776–4789. et al. (2007). Genetic variation in 1253 immune and inflammation Marafioti T, Jabri L, Pulford K, Brousset P, Mason DY, Delsol G. genes and risk of non-Hodgkin lymphoma. Blood 110: 4455–4463. (2003). Leucocyte-specific protein (LSP1) in malignant lymphoma Creighton CJ, Hilger AM, Murthy S, Rae JM, Chinnaiyan AM, and Hodgkin’s disease. Br J Haematol 120: 671–678. El-Ashry D. (2006). Activation of mitogen-activated protein kinase McInerney N, Colleran G, Rowan A, Walther A, Barclay E, Spain S in estrogen receptor alpha-positive breast cancer cells in vitro et al. (2009). Low penetrance breast cancer predisposition SNPs are induces an in vivo molecular phenotype of estrogen receptor alpha- site specific. Breast Cancer Res Treat 117: 151–159. negative human breast tumors. Cancer Res 66: 3903–3911. Meijers-Heijboer H, van den Ouweland A, Klijn J, Wasielewski M, de Dite GS, Jenkins MA, Southey MC, Hocking JS, Giles GG, McCredie Snoo A, Oldenburg R et al. (2002). Low-penetrance susceptibility to MR et al. (2003). Familial risks, early-onset breast cancer, and breast cancer due to CHEK2(*)1100delC in noncarriers of BRCA1 BRCA1 and BRCA2 germline mutations. J Natl Cancer Inst 95: or BRCA2 mutations. Nat Genet 31: 55–59. 448–457. Meyer KB, Maia AT, O’Reilly M, Teschendorff AE, Chin SF, Caldas Easton DF, Eeles RA. (2008). Genome-wide association studies in C et al. (2008). Allele-specific up-regulation of FGFR2 increases cancer. Hum Mol Genet 17: R109–R115. susceptibility to breast cancer. PLoS Biol 6: e108. Easton DF, Pooley KA, Dunning AM, Pharoah PD, Thompson D, Miki Y, Swensen J, Shattuck-Eidens D, Futreal PA, Harshman K, Ballinger DG et al. (2007). Genome-wide association study identifies Tavtigian S et al. (1994). A strong candidate for the breast and novel breast cancer susceptibility loci. Nature 447: 1087–1093. ovarian cancer susceptibility gene BRCA1. Science 266: 66–71. FitzGerald MG, Marsh DJ, Wahrer D, Bell D, Caron S, Shannon KE Moffa AB, Ethier SP. (2007). Differential signal transduction of et al. (1998). Germline mutations in PTEN are an infrequent cause alternatively spliced FGFR2 variants expressed in human mammary of genetic predisposition to breast cancer. Oncogene 17: 727–731. epithelial cells. J Cell Physiol 210: 720–731. Fu YP, Edvardsen H, Kaushiva A, Arhancet JP, Howe TM, Kohaar I Nordgard SH, Johansen FE, Alnaes GI, Naume B, Borresen-Dale AL, et al. (2010). NOTCH2 in breast cancer: association of SNP Kristensen VN. (2007). Genes harbouring susceptibility SNPs are rs11249433 with gene expression in ER-positive breast tumors differentially expressed in the breast cancer subtypes. Breast Cancer without TP53 mutations. Mol Cancer 9: 113. Res 9: 113. Garcia-Closas M, Chanock S. (2008). Genetic susceptibility loci for Orr N, Chanock S. (2008). Common genetic variation and human breast cancer by estrogen receptor status. Clin Cancer Res 14: 8000–8009. disease. Adv Genet 62: 1–32. Garcia-Closas M, Hall P, Nevanlinna H, Pooley K, Morrison J, Parkin DM, Bray F, Ferlay J, Pisani P. (2005). Global cancer statistics, Richesson DA et al. (2008). Heterogeneity of breast cancer 2002. CA Cancer J Clin 55: 74–108.

Oncogene GWAS in breast cancer D Fanale et al 2128 Peto J. (2002). Breast cancer susceptibility—A new look at an old independent gene expression data sets. Proc Natl Acad Sci USA model. Cancer Cell 1: 411–412. 100: 8418–8423. Peto J, Collins N, Barfoot R, Seal S, Warren W, Rahman N et al. Stacey SN, Manolescu A, Sulem P, Rafnar T, Gudmundsson J, (1999). Prevalence of BRCA1 and BRCA2 gene mutations in Gudjonsson SA et al. (2007). Common variants on chromosomes patients with early-onset breast cancer. J Natl Cancer Inst 91: 943–949. 2q35 and 16q12 confer susceptibility to estrogen receptor-positive Pharoah PD, Antoniou A, Bobrow M, Zimmern RL, Easton DF, breast cancer. Nat Genet 39: 865–869. Ponder BA. (2002). Polygenic susceptibility to breast cancer and Stacey SN, Manolescu A, Sulem P, Thorlacius S, Gudjonsson SA, implications for prevention. Nat Genet 31: 33–36. Jonsson GF et al. (2008). Common variants on chromosome 5p12 Rahman N, Seal S, Thompson D, Kelly P, Renwick A, Elliott A et al. confer susceptibility to estrogen receptor-positive breast cancer. Nat (2007). PALB2, which encodes a BRCA2-interacting protein, is a Genet 40: 703–706. breast cancer susceptibility gene. Nat Genet 39: 165–167. Tamaru N, Hishikawa Y, Ejima K, Nagasue N, Inoue S, Muramatsu Ricol D, Cappellen D, El Marjou A, Gil-Diez-de-Medina S, Girault M et al. (2004). Estrogen receptor-associated expression of JM, Yoshida T et al. (1999). Tumour suppressive properties of keratinocyte growth factor and its possible role in the inhibition fibroblast growth factor receptor 2-IIIb in human bladder cancer. of apoptosis in human breast cancer. Lab Invest 84: 1460–1471. Oncogene 18: 7234–7243. Tannheimer SL, Rehemtulla A, Ethier SP. (2000). Characterization Rosa-Rosa JM, Pita G, Urioste M, Llort G, Brunet J, Lazaro C et al. of fibroblast growth factor receptor 2 overexpression in the (2009). Genome-wide linkage scan reveals three putative breast- qhuman breast cancer cell line SUM-52PE. Breast Cancer Res 2: cancer-susceptibility loci. Am J Hum Genet 84: 115–122. 311–320. Ruiz-Narvaez EA, Rosenberg L, Rotimi CN, Cupples LA, Boggs DA, Thompson D, Duedal S, Kirner J, McGuffog L, Last J, Reiman A Adeyemo A et al. (2010). Genetic variants on chromosome 5p12 are et al. (2005). Cancer risks and mortality in heterozygous ATM associated with risk of breast cancer in African American women: mutation carriers. J Natl Cancer Inst 97: 813–822. the Black Women’s Health Study. Breast Cancer Res Treat 123: Thompson D, Easton D. (2004). The genetic epidemiology of breast 525–530. cancer genes. J Mammary Gland Biol Neoplasia 9: 221–236. Sidransky D, Tokino T, Helzlsouer K, Zehnbauer B, Rausch G, Wang X, Pankratz VS, Fredericksen Z, Tarrell R, Karaus M, Shelton B et al. (1992). Inherited p53 gene mutations in breast McGuffog L et al. (2010). Common variants associated with breast cancer. Cancer Res 52: 2984–2986. cancer in genome-wide association studies are modifiers of breast Smid M, Wang Y, Klijn JG, Sieuwerts AM, Zhang Y, Atkins D et al. cancer risk in BRCA1 and BRCA2 mutation carriers. Hum Mol (2006). Genes associated with breast cancer metastatic to bone. Genet 19: 2886–2897. J Clin Oncol 24: 2261–2267. Wooster R, Bignell G, Lancaster J, Swift S, Seal S, Mangion J et al. Sorlie T, Perou CM, Tibshirani R, Aas T, Geisler S, Johnsen H et al. (1995). Identification of the breast cancer susceptibility gene (2001). Gene expression patterns of breast carcinomas distinguish BRCA2. Nature 378: 789–792. tumor subclasses with clinical implications. Proc Natl Acad Sci USA Zhang Y, Gorry MC, Post JC, Ehrlich GD. (1999). Genomic 98: 10869–10874. organization of the human fibroblast growth factor receptor 2 Sorlie T, Tibshirani R, Parker J, Hastie T, Marron JS, Nobel A et al. (FGFR2) gene and comparative analysis of the human FGFR gene (2003). Repeated observation of breast tumor subtypes in family. Gene 230: 69–79.

Oncogene